BIOL 300
Spring 2014
Lab Manual for the Analysis of Biological Data
© 2008-2013 Michael Whitlock. Chapter 2 © Michael Whitlock, Brad Anholt, and
Darren Irwin
Biology 300 lab manual, Introduction
This book contains instructions for the lab exercises for BIOL300, the biostatistics
course at UBC. It is intended to complement, not to replace, the text Analysis of
Biological Data, by Whitlock and Schluter. The labs contain a mix of data
collection, computer simulation, and analysis of data using a computer program
called JMP (pronounced "jump").
All data described in these labs are real, taken from actual experiments and
observations reported in the scientific literature. References for each paper are
given in at the end of the manual.
This lab manual is a work-in-progress. Suggestions for improvements are
welcome.
This manual has been much improved by suggestions from Gwylim Blackburn,
Becca Gooding-Graf, Darren Irwin, Haley Kenyon, Heather Kharouba, Becca
Kordas, and Jess McKenzie.
Many of the labs use java applets, and these are to be found on the web. In order
to save you from typing those URLs, links are provided from the lab page at
http://www.zoology.ubc.ca/~irwin/BIOL300/labs/.
2
Biology 300 lab manual, Lab 1
1. Introduction to statistics on the computer and graphics
Goals

Get started on the computers, learning how to start with JMP

Collect a data set on ourselves for future use

Make graphs, such as histograms, bar charts, box plots, scatter plots, dot
plots, and mosaic plots.

Learn to graph data as the first step in data analysis
Quick summary from text (see Chapter 2 in Whitlock and Schluter)

Computers make data analysis faster and easier. However, it is still the
human's job to choose the right procedures.

Graphing data is an essential step in data analysis and presentation. The
human mind receives information much better visually than verbally or
mathematically.

Variables are either numerical (measured as numbers) or categorical
(describing which category an individual belongs to).

The distribution of categorical variables can be presented in a bar chart.
The distribution of numerical variables can be presented in a histogram, a
box plot, or cumulative frequency plot.

The relationship between two numerical variables can be shown in a
scatter plot. The relationship between two categorical variables can be
shown in a mosaic plot or a grouped bar chart. The association between a
numerical variable and a categorical variable can be shown with multiple
histograms, grouped cumulative frequency plots, or multiple box plots.

A good graph should be honest and easy to interpret, with as much
information as needed to interpret the graph readily available. At the same
time, the graph should be uncluttered and clear.
3
Biology 300 lab manual, Lab 1
Activities
1. Log on to the computer with your new password.
(See the next section if you need detailed instructions.) Remember this
username and password; you'll need them throughout the term.
2. Recording data.
Using the "Student data sheet 1" at the end of this manual, record the
requested information about yourself. This is optional; if you have any
reason to not want to record this (relatively innocuous) data about
yourself, you do not have to. If you feel that you would like to skip just one
of the bits of information and fill in the rest, that is fine too. The data
sheets do not identify students by name. Pass the sheet to the instructor
when you are finished.
Learning the Tools
1. How to log on
a. If the screen is dark, make sure the screen is turned on, with the
lighted button in the bottom right corner. Move the mouse or hit the
shift key to wake up the computer.
b. You may be prompted to hit Ctrl-Alt-Delete (three buttons all at the
same time). Hit OK to the next screen.
c. Enter the username and password into the appropriate places, from
the slip of paper given to you by the TA. Keep track of these; you'll be
using these throughout the term.
d. After the computer starts up, double-click on the icon for "JMP". This is
the statistical package we'll use most in this class.
e. After you're done for the day, don’t forget to log-off using the menu in
the lower left corner of the screen and hit OK.
2. Opening a data file
Using the JMP starter, click on "Open data table." This lets you open
previously saved files.
Navigate to the shared drive called "Shared on 'Statslab Server
(Statslab)'". This is where all the previously collected data sets needed for
this class are stored. You can also create and store new data sets or
copies of these others in your own folders with your account.
4
Biology 300 lab manual, Lab 1
Open a file called "titanic.csv". This file has information on all the
passengers of the RMS Titanic on its initial and disastrous run.
3. Check that the variables are labeled as “continuous” or “nominal”
correctly.
The
and
icons that you see to the left of your data file tell you
what the computer thinks is the type of each of your variable. The
icon
denotes a numerical variable (what JMP calls “Continuous”), and the
icon makes categorical variables (what JMP calls “Nominal”). If the type is
incorrect, just click on the
or
from the menu that appears.
icon and choose the correct type
4. Changing column information.
You can change information about the column (e.g. its title, the data type,
etc.) by double-clicking on the column heading. A straightforward menu
will appear.
5. Making a graph of one variable
For example, using the Titanic data in
"titanic.csv", let's plot a histogram of the
ages of the passengers. From the
menu bar, click "Analyze" and under
that menu click "Distribution". A window
will appear. In that window click on
"Age" under "Select columns", then
click on the button labeled "Y,
Distributions" under the second column,
and then click OK.
5
Biology 300 lab manual, Lab 1
A window will appear that includes the histogram, although it will be turned
on its side. (If you want the histogram to appear in the more typical
horizontal layout, click on the red triangle  next to the variable name
"Age" to open a menu, click "Display Options" and then "Horizontal
Layout".)
We can plot a bar chart showing the distribution of a categorical variable in
exactly the same way, for example with the variable "Passenger class."
(Note: JMP makes bar charts without spaces between the bars, which is a
flaw in the program.)
6. Making a graph of two or more variables
To plot the relationship between two variables, choose "Fit Y by X" from
the "Analyze" menu bar listing.
In the window that opens, select the explanatory variable you are
interested in from the list on the left, and then click the button labeled "X,
Factor" from the middle column. For example, click on "Sex" and then click
"X, Factor." Then click on the response variable from the list on the left,
and then click the button labeled "Y, Response". Finally click "OK".
A window will appear that shows an appropriate graph of the relationship
between the two variables. JMP will choose a graphic type that matches
the variable types of the selected variables, so the same procedure will
give a mosaic plot for two categorical variables, a scatter plot for two
numerical variables, or dot plots for one categorical and one numerical
variable.
6
Biology 300 lab manual, Lab 1
7. Making graphs for more than one group at a time.
In either the "Fit Y by X" or "Distribution" menus, you can plot graphs of
subsets of the data by clicking on one of the variables and clicking the
"By" button. For example, to plot histograms of height for each sex
separately, put "Age" in the "Y, Distribution" box of the "Distribution"
window, and put "Sex" in the "By" box. JMP will give you separate results
for males and females.
8. Using the manual.
These course notes will only scratch the surface of the possible
techniques available in JMP. You can find out about more features by
using the online manuals. The search function on JMP help is substandard, so you may find it easiest to look in the index of one of the
books available online, such as the Statistics and Graphics Guide. Look
under the "Help" menu for "Books" and choose the "JMP Stat and Graph
Guide".
9. JMP tips
At the end of this lab manual, there are a couple of pages with some
general tips on how to use JMP more easily and powerfully.
10. Box plots
JMP automatically graphs numerical data using box plots, typically shown
to the right of the histogram. Below is the box plot for the variable
"Mortality rate adult female" from the "countries2005.csv" file. You will
learn about the various things the box plot shows later (e.g. lab 3), but we
show it now to increase familiarity with the graphing output of JMP.
Essentially, the box plot illustrates at a glance where the bulk of the
distribution is (in the box) and the range of the data, as well as other
details. We've added labels of the various parts:
7
Biology 300 lab manual, Lab 1
Questions
1. For each of the following pairs of graphs, identify features that are better on
one version that the other.
a. Survivorship as a function of sex for passengers of the RMS Titanic
b. Ear length in male humans as a function of age
2. Open the data file called "countries2005." This file gives various data from the
World Bank on all countries of the world for 2005.
a. Plot a histogram for the variable "birth rate crude (per 1000 people)."
b. Look carefully at the histogram. Do you see evidence of a problem with
the data file? Find and fix the problem.
(Two practical hints: 1. An individual can be identified from a plot by
clicking on a dot or bar of the histogram. This will select the row or
rows associated with that part of the graph. 2. A row can be excluded
from analysis by selecting the row and then choosing "Delete Rows"
from the "Rows" menu.)
c. Plot the histogram on the corrected data set.
8
Biology 300 lab manual, Lab 1
3. Muchala (2006) measured the length of the tongues of eleven different species
of South American bats, as well as the length of their palates (to get an indication
of the size of their mouths). All of these bats feed on nectar from flowers, using
their tongues. Data from their paper are given in the file "battongues.csv."
a. Plot a scatter plot with palate length as the explanatory variable and
tongue length as the response variable. Describe what you see.
b. All of the data points in this graph have been double checked and
verified. With that in mind, how do you interpret the outlier on the
scatter plot? (You might want to read about these bats in the Muchhala
(2006) paper, to learn more about the biology behind Anoura fistulata.)
4. Use the countries data (as corrected as per Question 2). Plot distributions for
"Continent," "Prevalence of HIV," and "Physicians per 1000 people." What kinds
of variables are these (numerical or categorical)? What kinds of graphs are being
drawn?
5. Use the countries data (as corrected in Question 2). Plot the relationship
between the following sets of variables:
a. Male life expectancy and female life expectancy,
b. Continent and life expectancy,
c. Literacy rates and life expectancy
d. Personal computers and life expectancy
e. Number of physicians and life expectancy
f. Which variable seems to explain life expectancy better: number of
personal computers or number of physicians? Try to explain the
pattern that you see.
9
Biology 300 lab manual, Lab 1
6. Use the data set collected on your class from today. (We'll return later to some
of the other variables later in the term.) Plot the relationship between the
following pairs of variables. For each case describe the pattern that you observe:
a. Handedness and "footedness"
b. Handedness and dominant eye
c. Sex and height
d. Height and head circumference
7. Use the data from class today. Plot the distributions of height for males and
females separately.
8. Pick one of the plots made by the computer program. What could be improved
about this graph to make it a more effective presentation of the data?
9. (Optional) -- This last exercise will teach some slightly more advanced uses
techniques for using JMP.
a. JMP will do other kinds of plot as well. For example, for the distribution
of a numerical variable, you can plot Cumulative Frequency
Distributions (CDF) or stem-and-leaf plots. To do so, get to the results
window of the Distribution process, and then under the red triangle 
click on "CDF Plot" or "Stem and Leaf."
b. Many features on JMP are quite intuitive, especially once you know
about the tiny red triangles . Play around with the program and a
data set to see what you can discover.
10
Biology 300 lab manual, Lab 2
2. Sampling
Goals

Explore the importance of random sampling

Investigate sampling error and sampling bias

Learn to create data files.
Quick summary from text (Chpt. 1 in Whitlock and Schluter)

With random sampling, all individuals in the population have an equal and
independent chance of being selected for the sample.

Most estimation methods and hypothesis tests assume that the data are a
random sample from the population.

Sampling error is the chance difference between an estimate describing a
sample and that parameter of the whole population. (Note that sampling
error is not due to measurement error or sampling bias, but rather arises
naturally from the process of random sampling.)

If individuals are not taken independently of each other, then the sample is
effectively smaller than it seems. The resulting estimates will have higher
sampling error than those taken from the same size random sample.

If some individuals are more likely than others to be selected for the
sample, then the sample has a high probability of being biased. If so, we
say that the process is subject to sampling bias.

Estimates are made from data. The goal of an estimate is to give a close
idea of the true value of a parameter in a population.

Estimates are almost always wrong, in the sense that they rarely exactly
match the true value of the parameter. Estimates can be biased, meaning
that on average multiple estimate taken in the same way will be different
from the true value. Estimates can also be imprecise, meaning that
different samples will give different estimated values.
11
Biology 300 lab manual, Lab 2
Learning the tools
In most computer programs for data analysis, data is entered in a particular way
into a table, or spreadsheet. These spreadsheets are divided into lots of small
rectangles called cells, and each cell belongs to a column (a vertical stack of
cells) and a row (a horizontal set of cells). The convention is that each row
corresponds to all the data from a single individual, and each column gives the
data for all individuals for a single variable.
Key point: In a stats spreadsheet like JMP, each row is a distinct individual, and
each column is a variable.
For example, here is a part of a data file on the passengers of the Titanic, which
shows the rows for 10 individuals, and for each individual, their values for seven
different variable: passenger class, name, age, the place they embarked from,
their destination, their sex, and whether they lived or died during the disaster,
with a yes or no. The full data set includes many more rows, for all the other
passengers aboard the ship. In the first row, for example, we see that a female
passenger named Miss Elisabeth Walton Allen traveled in first class from
Southampton to St. Louis, and she survived the journey.
Setting up a new data file is an extremely important part of data analysis. If
variables are not labeled clearly and unambiguously, the results will not be
interpretable later.
1. Open a new file.
In JMP, click on the "New Data Table" button from the JMP starter. Save it
as "disaster.jmp." (Note -- we're not going to use this new file; this is just
for an example of how to do the steps.)
2. Make new columns.
For each variable in your data, add a new column. Under the menu
labeled "Cols" choose "Add multiple columns." Enter the number of
variables in the data, in this case 7, to the box labeled "How many
columns to add." Click "OK."
12
Biology 300 lab manual, Lab 2
For each column, change the label to the desired variable name and tell
the computer the type of variable it is. For example, for the first column,
double click on the white space at the top of the column. A new window
will open which allows you edit the column name (put "Passenger Class").
Also, this data is not numerical but is based on alphabetic characters. So
click on the "Data type" window and click on "character." "Modeling type"
should also change to "nominal" as a result. These changes tell the
computer to treat the data in this column as categorical data, not as
numerical data. This is an essential step in having the analysis work
correctly later.
Finish this with other columns, if you like. Name, embarked, destination,
sex, and survived are categorical variables, while age is a numerical
variable.
3. Enter the data.
For each individual, add a row and type in the values for that individual
into the appropriate cells. The "Tab" key will take you easily to the next
cell. JMP will automatically start a new row if you hit "Tab" in the last cell
of the previous row.
4. Summarizing numerical data
To find the mean of a numerical variable, choose "Distribution" from the
"Analyze" menu. Click on the variable name you are interested in (for
example Age), then click on "Y, columns" and OK. The mean of the
variable will appear among a list of summaries of the data below the
graph.
13
Biology 300 lab manual, Lab 2
5. Creating new columns using other variables.
Sometimes it can be useful to create new variables that use other data
already in the data file. For example, we might want to convert a value
measured in inches to one measured in centimeters. We might want to
take the log of each data point. Or, as we will need later, we might want to
calculate the difference between two variables.
As an example, let's make a column with approximate age in months. To
calculate the new variable, first create a new column in the same way as
above ("Cols" > "New Column..."). Name the new variable in the resulting
box, and choose the correct data type for the new variable. In this case,
let's make a numerical variable called "age (months)".
After returning to the data window, click on any cell in the new column,
and then click on "Formula..." under the "Cols" menu. A new window will
open. This window contains a space at the bottom to create a formula that
uses other data and other mathematical functions. To make our new
column, double click on "Age" to add it to the bottom window, then click on
the multiplication sign, then enter 12 to the new box and hit return. When
finished building the formula, hit the "OK" button.
14
Biology 300 lab manual, Lab 2
Activities
In this class you will be measuring several cowrie shells and then comparing the
measurements by using the computer to plot the data.
The class divides itself into groups of two to four people. Each group will be given
a container of cowrie shells and a white board. Follow these steps:
1. As a group, sample five shells from this container. Put them in spots 1 through
5 on the white board. This is your sample.
2. For each shell that you have sampled, measure the length of the shell, and
write these lengths down on the white board by the shell.
3. Open a new data file in JMP, and enter the lengths of your five shells. Save
this with a title something like “First sample.jmp”.
4. Calculate the mean and standard deviation of the length of these five shells.
Record this mean.
5. Next we will calculate the mean of the population of shells in your container.
Take each remaining shell one by one, measure its length, and record it on the
whiteboard beside the next available number. Set the shell there on the
whiteboard as well.
6. Open another data file in JMP. Add the lengths of all the shells into this file
under a column for length. Save this as something like “Full population
shells.jmp”.
7. Calculate the mean and standard deviation of the full population.
8. Is the mean of your sample the same as the mean of the population as a
whole? What about the standard deviation? Explain reasons why the sample
mean and the population mean could differ.
9. Let’s now explore how to randomly sample from the population. You should
have each shell now next to a unique number on your whiteboard. To randomly
choose five of these for a random sample, we can use random numbers
generated by http://www.random.org/sequences/. Open this web page, tell it that
the “smallest value” is 1 and the “largest value” is the number of shells in your
population. Click “Get sequence”. Use the first five numbers in the list to tell you
which shells to sample.
10. We’re going to randomly sample from the population several times. Each
sample is going to have five shells, and we’ll calculate the mean of each sample.
Open a new file in JMP, and give it two columns. The first column will tell us
which sample an individual belongs to, and this should be specified as a nominal
15
Biology 300 lab manual, Lab 2
(categorical) variable called “Sample”. The second column will be the length
measurement for a randomly chosen shell, so this column should be specified as
a “continuous” (i.e. numerical) variable, called “Length.”
11. For the first sample, put “S1” in each of the first five rows of this file in the
“Sample” column. “S1” will indicate that an individual was part of “sample 1.” This
sample will have five individuals in it, so we need five rows. For the second
sample, you’ll want five rows with “S2” for the label of the sample, etc., in the
same file.
12. Record the length of the five shells in your first random sample, in the second
column, one length for each row.
13. Repeat steps 11 and 12 thirty times, changing the “Sample” variable
appropriately (“S1” through “S30”). Ask the random number generator for a new
set of random numbers each time to choose which shells to use for each sample.
14. Create a new file that has the mean shell lengths for each of the randomly
chosen samples. You can do this with the “Tables” menu bar, and then select
“Summary”. In the resulting window, put the name of your sample label column in
the box for Groups. The select the name of the length column, and then click
“Mean” in the pop-up menu that appears when you click “Statistics”. When you
hit “OK”, JMP should create a new data file with the means of each sample.
Questions
1. Plot a histogram in JMP that shows the distribution of the means of your
random samples. Describe the shape of this distribution. Does every sample
return the same value of the mean? Why, or why not?
16
Biology 300 lab manual, Lab 2
2. What is the mean of the distribution of the random sample means? Is it close
to the mean shell length of the population?
3. How does the mean of your very first sample (in step 1) compare to the
population mean? Is it more or less different from the population mean than a
typical sample mean from the randomized samples? If so, why might that be?
Assignment (not to be done in Spring 2014, but good to think
about anyway)
Write a report documenting the results of your sampling exercises this week.
Address these two questions: "Is there evidence of sampling bias in your fist
sample?" and " How much does sampling error affect the reliability of using a
sample of five individuals to estimate the mean shell length of a population?"
Write the report in the format of a scientific paper, with a brief introduction
describing the questions, a methods section (which needs to specify your
sampling system as well as the basic approach detailed above), results, and
discussion. The report should not require more than 3-4 pages of double-spaced
typed text. More advice on writing a good report is given at the end of this lab
manual.
17
Biology 300 lab manual, Lab 3
3. Describing data and estimation
Goals

Investigate sampling error; see that larger samples have less sampling
error.

Visualize confidence intervals.

Calculate basic summary statistics using the computer.

Calculate confidence intervals for the mean on the computer.

Understand box plots.
Quick summary from text (Chpts. 3 and 4 in Whitlock and Schluter)

The mean describes the "center of gravity" of a set of numbers. The
median is the data point that is smaller than half of the data and greater
than the other half of the data.

The standard deviation is a description of the typical spread of data
around the mean. The standard deviation is the square root of the
variance, which is the average squared deviation from the mean.

The 3rd quartile is the data point greater than three quarters of the other
data. The 1st quartile is the point greater than one-quarter of the other
data. The interquartile range is the difference between the third and first
quartiles. The interquartile range is another measure of the spread of a set
of data.

The skew of a set of data describes the asymmetry of the distribution.
Zero skew means that the distribution is symmetric. If the skewness is
positive, then the distribution tends to have a long tail to the right. With
negative skewness, the distribution has a long tail to the left.

Sampling error is the variation in estimates among samples caused by
chance deviations between the sample and population.

Sampling error is greater in small samples than in large samples. Large
samples are more likely to give a precise estimate.

Sampling error is measured by the standard error. The standard error is
the standard deviation of the sampling distribution of an estimate. The
standard error of a mean can be calculated by the standard deviation
divided by the square root of the sample size: SEY = s / n .
18
Biology 300 lab manual, Lab 3

The reliability of an estimate can be expressed as a confidence interval. a
95% confidence interval will contain the true value of the parameter in
95% of estimates.
Activities
Distribution of sample means
Go to the web and open the page at
http://onlinestatbook.com/stat_sim/sampling_dist/index.html. This page contains
a java applet that lets you play around with sampling, to see the distribution of
sampling means. After the page loads, click the "Begin" button. You'll see a
window that looks like this:
The top box contains a distribution that represents the true distribution of a
variable in the population. This is the true universe that the program is using to
sample individual data points from. In the example pictured here, the mean of the
population is 16, with standard deviation 5, and the individuals in the population
occur according to a normal (bell-shaped) distribution.
Next click the "Animated" button on the right side. This will draw five individuals
at random from the population, and plot a histogram of the sample data in the
second graph. It will also calculate the mean of the sample, and start to create a
histogram in the third graph. The third graph shows the distribution not of
individuals, but of sample means. Notice how the bar added in the distribution of
sample means corresponds to the mean of the sample in the second graph.
If you click "Animated" again, it will draw an entirely new sample from the
population, and then add the mean of that new sample to the distribution of
sample means. Do this a few times, and watch what's happening.
19
Biology 300 lab manual, Lab 3
After you get bored of that, click the button that says "5". This button will draw 5
different samples, and add the five new sample means to the histogram of
sample means. The "1,000" and "10,000" buttons do the same thing, except for
more samples.
After adding a large number of samples, look at the distribution of sample means.
What kind of shape does it have? It ought to look normal as well. Also look at the
width of the distribution of sample means - is it more variable, less variable or
about as variable as the population distribution? What do you expect?
Next, hit the "Clear lower 3" button at the top. This will erase all the distributions
that you just drew. Use the "N=5" button to choose a new sample size, starting
with N=25. Repeat the steps above to generate a distribution of sample means.
Now the sample means are based on 25 individuals instead of just 5 individuals.
Are the sample means more or less variable with N=25 than with N=5?
Confidence intervals
Go back to the web, and open the java applet at
http://onlinestatbook.com/stat_sim/conf_interval/index.html. This applet draws
100 different confidence intervals for the mean of a known population. Hit
"Sample," and look at the results. Each horizontal line on the graph represents a
different sample of 10 individuals (or more, if you change the sample size). Each
line shows the 95% confidence interval for the mean (the inside part of the line in
red or yellow) as well as the 99% confidence interval (the wider line in blue or
white).
The 95% confidence interval is shown in yellow if the interval contains the true
value of the mean, represented by the vertical line down the screen. The
program changes the 95% confidence interval line to red if the 95% confidence
interval fails to enclose to true mean. Out of the 100 samples, how many 95%
confidence intervals contain the true mean? How many missed?
Click "Sample" a few more times. Each time you click, the program will draw 100
more samples and draw them. It also keeps track of all the sample you have
drawn in the table at the bottom. After drawing a lot of samples (say 1000), what
proportion of samples have 95% confidence intervals that include the true mean?
What proportion fail to enclose the true mean?
What about the 99% confidence interval -- what fraction of 99% confidence
intervals enclose the true mean?
A 95% confidence interval ought to enclose the true value of the parameter for 95
samples out of 100, on average, and the 99% confidence interval ought to
succeed 99% of samples. In fact, of course, that is the definition of a 95% (or
99%) confidence interval; the percentage tells us the expected probability that
the true mean is captured within the confidence interval.
20
Biology 300 lab manual, Lab 3
Learning the Tools
1. Calculating summary statistics
a. Use the "Distribution" window, found under the "Analyze" menu bar. In
the window that opens, choose the variable (or variables -- you can do
more than one variable at once), click "Y-columns", and hit "OK." A
graph will appear, and below the graph will be a collection of summary
statistics including mean, standard deviation, standard error of the
mean, and the 95% confidence interval for the mean.
b. To find a longer list of summary statistics, click on the red triangle 
next to the variable name. From the resulting menu, click on "Display
Options" and then "More moments." This will add to the list of summary
statistics, including the variance, skewness, and the coefficient of
variation (CV).
Key point: The red triangles  sprinkled through JMP windows open new menus
of options when clicked.
2. Calculating confidence intervals of means
The 95% confidence interval is automatically calculated by opening the
"Distribution" window for a variable. For other confidence intervals, open
the "Distribution" window, and the click the red triangle  next to the
variable name. Select "Confidence Intervals" and choose the desired
confidence level from the list. (For example, for a 99% confidence interval,
choose "0.99"). This will add a section to the results with the confidence
intervals for the mean and standard deviation of the variable.
21
Biology 300 lab manual, Lab 3
Questions
1. The data file "students.csv" should include data that your class collected on
themselves during the first week of lab, in chapter 1. Open that file.
a. What is the mean height of all students in the class?
b. Look at the distribution of heights in the class. Describe the shape
of the distribution. Is it symmetric or skewed? Is it unimodal or
bimodal?
c. Calculate the standard deviation of height.
d. Calculate the standard error of height. Does this match the
SEY = s / n that you expect?
e. Calculate the mean height separately for each sex. Which is
higher? Compare the standard deviations of the two sexes as well.
2. The file "caffeine.csv" contains data on the amount of
caffeine in a 16 oz. cup of coffee obtained from various
vendors. For context, doses of caffeine over 25 mg are
enough to increase anxiety in some people, and doses
over 300 to 360 mg are enough to significantly increase
heart rate in most people. Red Bull contains 80mg of
caffeine per serving.
a. What is the mean amount of caffeine in 16 oz.
coffees?
b. What is the 95% confidence interval for the
mean?
c. Plot the distribution of caffeine level for these
data. Is the amount of caffeine relatively
predictable in a cup of coffee? What is the standard deviation of
caffeine level? What is the coefficient of variation?
d. The file "caffeine-Starbucks.csv" has data on six 16 oz. cups of coffee
sampled on six different days from a Starbucks location, for the
Breakfast Blend variety. Calculate the mean (and its 95% confidence
interval) for these data. Compare these results to the data taken on the
broader sample in the first file, and describe the difference.
3. A confidence interval gives a range of values that are likely to contain the true
value for a parameter. Consider the "caffeine.csv" data again.
22
Biology 300 lab manual, Lab 3
a. Calculate the 95% confidence interval for the mean caffeine level.
b. Calculate the 99% confidence interval for the mean caffeine level.
c. Which is larger (i.e. has a broader interval)? Why should this one be
larger?
d. Calculate the quantiles of the distribution of caffeine. (Big hint: they
automatically appear in the "Distribution " window, immediately below
the graphs.) What are the 2.5% and 97.5% quantiles of the distribution
of caffeine? Are these the same as the boundaries of the 95%
confidence interval? If not, why not? Which should bound a smaller
region, the quantile or the confidence interval of the mean?
4. Return to the class data set, "students.csv." Find the mean value of "number of
siblings." Add one to this to find the mean number of children per family in the
class.
a. The mean number of offspring per family twenty years ago was about
2. Is the value for this class similar, greater, or smaller? If different,
think of reasons for the difference.
b. Are the families represented in this class systematically different from
the population at large? Is there a potential sampling bias?
c. Consider the way in which the data was collected. How many families
with zero children are represented? Why? What effect does this have
on the estimated mean family size?
5. Return to the data on countries of the world, in "countries2005.csv" and the
data from the class in "students.csv." Plot the distributions and calculate
summary statistics for land area, life expectancy (total), personal computers per
100 people, physicians (per 1000 people), as well as height for each sex.
Key point: A distribution is skewed if it is asymmetric. A distribution is skewed
right if there is a long tail to the right, and skewed left if there is a long tail to the
left.
a. For each variable, plot a histogram of the distribution. Is the variable
skewed? If so, in which direction?
23
Biology 300 lab manual, Lab 3
b. For each variable, calculate the mean and median. Are the similar?
Match the difference in mean and median to the direction of skew on
the histogram. Do you see a pattern?
6. A box plot is a graphical technique to show some key features of the
distribution of a numerical variable. Unfortunately, box plots are not always drawn
in exactly the same way. JMP draws box plots automatically with the
"Distribution" window, to the right of the histogram. Plot the box plot for "Mortality
rate adult female" from the "countries2005.csv" file. It should look like this, except
that we've added labels of the various parts:
The interquartile range is the difference between the 75th percentile and the 25th
percentile. It is an alternate measure of the spread of the distribution.
Look at the values in the Quantiles and Moments sections below the graphs, and
read them off this graph. Confirm that these designations on this graph are
correct.
24
Biology 300 lab manual, Lab 4
4. Probability
Goals

Practice making probability calculations

Investigate probability distributions
Quick summary from text (Chpt 5 in Whitlock and Schluter)

The addition principle says that the probability of either of two mutually
exclusive events is the probability of the first event plus the probability of
the second event: Pr[A or B] = Pr[A] + Pr[B] .

The general addition principle says that the probability of either of two
events is the probability of the first event plus the probability of the second
event, minus the probability of getting both events: Pr[A or B] = Pr[A] +
Pr[B] − Pr[A and B].

The multiplication principle says that the probability of two events both
occurring - if the two events are independent -- is the probability of the first
times the probability of the second: Pr[A and B] = Pr[A] Pr[B].

The general multiplication rule says that the probability of two events both
occurring is the probability of the first event times the probability of the
second event give the first: Pr[A and B] = Pr[A] Pr[B | A].

A probability tree is a graphical device for calculating the probabilities of
combinations of events.

The law of total probability, Pr [ A ] = ∑ Pr [ B ] Pr [ A | B ] , makes it possible to
all B
calculate the probability of an event (A) from all of the conditional
probabilities of that event. The law adds up for all possible conditions (B)
the probability of that condition (Pr[B]) times the conditional probability of
the event assuming that condition (Pr[A | B]).

The binomial distribution describes the probability of a given number of
successes out of n trials, where each trial independently has a p
! n $ X
n−X
probability of success: Pr [ X ] = #
& p (1− p) .
" X %
25
Biology 300 lab manual, Lab 4
Activities
1. Let's Make a Deal
An old TV game show called "Let's Make a Deal" featured a host, Monty
Hall, who would offer various deals to members of the audience. In one
such game, he presented a player with three doors. Behind one door was
a fabulous prize (say, a giant pile of money), but behind the other two
were relatively worthless gag gifts (say, donkeys).
The player was invited to choose one of the doors. Afterwards, the host
would open one of the other two doors to reveal that it contained a
donkey. Then, the player is offered a choice: they can either stay with their
original door choice, or they can switch to the other closed door.
The question becomes, which strategy is better -- to stick with the original
or to switch doors? In the last few years, this relatively simple question
has attracted a lot of debate, with professional mathematicians getting
different answers.
a. Find the probability of finding the door with the pile of money, if the
player follows the strategy of staying with their original choice.
b. Find the probability of success with strategy of switching to the other
closed door.
c. Using the simulation at http://www.stayorswitch.com, play the game
20-30 times using each of the two strategies. Was your answer
correct? Remember that because the game in the simulation is a
random process, the proportion of successes will not necessarily
exactly match your theoretical predictions. (Note that there is a version
of the game at another website that will keep track of your results,
although it is a little clunky to use:
http://www.stat.tamu.edu/~west/applets/LetsMakeaDeal.html
2. The binomial distribution
This is a simple, quick exercise meant to help you visualize the meaning
of a binomial distribution. The binomial distribution applies to cases where
we do a set number of trials, and for each trial there is an equal and
independent probability of a success. Let's say there are n trials and the
probability of success is p. Of those n trials, anywhere between 0 and n of
them could be a 'success" and the rest will be "failures." Let's call the
number of successes X. This is relevant to many biological situations; for
example, calculating the probability of getting exactly three daughters in a
family of five, when you assume a certain fixed probability of each baby
being female.
26
Biology 300 lab manual, Lab 4
As our example, let's roll five regular dice, and keep track of how many
come up as "three." In this case, there are five trials, so n = 5. The
probability of rolling a "three" is 1/6, so p = 0.1666.
Let's repeat the sampling process a lot of times. For each sample, roll five
dice, and record the number of threes.
For this exercise, let’s draw a histogram of the results by hand. First draw
the axes, and then for each result draw an X stacked up in the appropriate
place. For example, after three trials that got 0, 1, and 1 threes,
respectively, the histogram will look like this:
Keep doing this until you have 20 or more samples.
Here is the frequency distribution for this process as calculated from the
binomial distribution.
Is this similar to what you found? (Remember that with only a few trials,
you don’t expect to exactly match the predicted distribution. If you
increase your number of samples, the fit should improve.)
The expected mean number of threes is 0.833. (For the binomial
distribution, the mean is n p, or in this case 5 × 0.1666.) Calculate the
mean number of threes in your samples.
27
Biology 300 lab manual, Lab 5
5. Frequency data and proportions
Goals

Better understand the logic of hypothesis testing.

Make hypothesis tests and make estimates about proportions.

Fit frequency data to a model.
Quick summary from text (Chpts 7 + 8 from Whitlock and Schluter)

The null hypothesis (H0) is a specific claim about a parameter. The null
hypothesis is the default hypothesis, the one assumed to be true unless
the data lead us to reject it. A good null hypothesis would be interesting to
reject.

The alternative hypothesis (HA) includes all values for the parameter other
than that stated in the null hypothesis.

The confidence interval for a proportion is best calculated by the AgrestiCoull method. Most statistical packages on the computer do not calculate
this method.

A binomial test is a hypothesis test that compares a hypothesized value of
a proportion to count data.

Categorical data describes which category an individual belongs to.
Categorical data can be summarized by the frequency or count of
individuals belonging to each category.

In some cases, the frequency of each category can be predicted by a null
hypothesis. We can test such null hypotheses with Goodness-of-Fit tests.

A common Goodness-of-Fit test is the χ2 test, which compares the number
of individuals observed in each category to the number expected by a null
hypothesis.

The χ2 Goodness-of-Fit test assumes that each individual is sampled
randomly and independently.

The χ2 test is an approximate test that requires that the expected numbers
in each category is not less than one, and that no more than 20% of the
categories have expected values less than 5.
28
Biology 300 lab manual, Lab 5

The χ2 test can be used as a quick approximation to the binomial test,
provided that the expected values of both categories are 5 or greater.

The Poisson distribution predicts the number of successes per unit time or
space, assuming that all successes are independent of each other and
happen with equal probability over space or time.
Activity
We'll do an experiment on ourselves. The point of the experiment needs to
remain obscure until after the data is collected, so as to not bias the data
collection process.
After this paragraph, we reprint the last paragraph of Darwin's Origin of
Species. Please read through this paragraph, and circle every letter "t".
Please proceed at a normal reading speed. If you ever realize that you
missed a "t" in a previous word, do not retrace your steps to encircle the
"t". You are not expected to get every "t", so don't slow down your reading
to get the "t"s.
k
It is interesting to contemplate an entangled bank, clothed with many plants of
many kinds, with birds singing on the bushes, with various insects flitting
about, and with worms crawling through the damp earth, and to reflect that
these elaborately constructed forms, so different from each other, and
dependent on each other in so complex a manner, have all been produced by
laws acting around us. These laws, taken in the largest sense, being Growth
with Reproduction; inheritance which is almost implied by reproduction;
Variability from the indirect and direct action of the external conditions of life,
and from use and disuse; a Ratio of Increase so high as to lead to a Struggle for
Life, and as a consequence to Natural Selection, entailing Divergence of
Character and the Extinction of less-improved forms. Thus, from the war of
nature, from famine and death, the most exalted object which we are capable
of conceiving, namely, the production of the higher animals, directly follows.
There is grandeur in this view of life, with its several powers, having been
originally breathed into a few forms or into one; and that, whilst this planet
has gone cycling on according to the fixed law of gravity, from so simple a
beginning endless forms most beautiful and most wonderful have been, and are
being, evolved.
k
Problem number 6 will return to this exercise. Please don't read Problem 5
until you have completed this procedure.
Learning the Tools
1. Binomial tests
29
Biology 300 lab manual, Lab 5
A binomial test compares the proportion of individuals in a sample that
have some quality to a hypothesis about that proportion.
In JMP, binomial tests can be made from the "Distribution" window. From
the menu bar, choose "Analyze", then "Distribution" and then choose the
variable that you want to study and hit OK. For example, the data set
called "bumpus.csv" contains data on a sparrow population caught in a
wind storm in the 1880's. These data were among the first to demonstrate
natural selection in a wild population. Use "Distribution" for the "sex"
variable.
In the Distribution window, choose "Test Probabilities" from the menu that
opens from the red triangle by the variable name. This expands the
results window below, and give a set of boxes to add the hypothesized
values for the proportions. Enter the null hypothesis values there; for
example, we could ask whether the Bumpus data has equal
representation of males and females, so we add 0.5 and 0.5 to the two
boxes.
Next, click one of the lines that say "Binomial test." Click "Done", and the
results appear.
2. Using "Frequency"
There are two ways that JMP can take frequency data. The first we have
already seen: each row represents an individual, and each individual has
a column with a categorical variable that describes which group that
individual belongs to.
Alternatively, in some cases it can be quicker to use a different format. If
there are multiple individuals that are identical (for example, if we had a
data set that recorded sex only, the rows would either be male or female
only). In these cases, we can create a row that describes those
individuals, and then add a column that records how many individuals are
like that -- the frequency. For example, the sex of the sparrows could be in
a data file that looked like this:
30
Biology 300 lab manual, Lab 5
In case like this, when we start to analyze the data, we need to tell the
program which column has this frequency information. We can do this in
either the "Distribution" or "Fit Y by X" menus, but adding the column with
frequency data to the blank for "Frequency".
3. Goodness of fit tests
A χ2 goodness of fit test can also be done from the "Distribution" window.
Follow the instructions for a binomial test to choose "Test Probabilities"
from the red triangle menu. After entering the expected proportions for
each category, click on "probabilities not equal to hypothesized values
(two-sided chi-square test)". (Note: for JMP, the hypothesized (or
expected) values can be entered either as proportions or counts. In the
results listing, the χ2 results will be shown in the row labeled "Pearson"
(after the inventor of the χ2 goodness of fit test, Karl Pearson).
For example, fitting the male/female data from the Bumpus data set to a
50:50 model, the P-value from the χ2 test is 0.0011. Note that when JMP
says “Prob>Chisq” in the bottom right corner of this box, this is telling you
the P-value of the test, or 0.0011 for Pearson’s χ2 test. You don’t need to
look up the χ2 value in a statistical table; JMP has already found the Pvalue for you.
31
Biology 300 lab manual, Lab 5
A goodness of fit test can be done using the "Frequency" option described
in point 2, as well.
4. Testing fit to a Poisson distribution
JMP can be used to test the goodness of fit to specific probability
distributions. For example, we may want to compare a set of data to the
expectations of a Poisson distribution. In this case, the null hypothesis is
that the data are drawn from a Poisson distribution.
Open the file “cardiac events.csv”. This file records the number of heart
attacks occurring per week in a hospital. The column titled “Number of
events per week” is the data showing number of heart attacks in a given
week. The column “Frequency - in-hospital cardiac events” shows the
number of weeks that had that number of events. For example, there were
ten weeks during this time period in which there were exactly five out-ofhospital cardiac events. There are a total of 261 weeks in the data set.
To ask whether these data fit a Poisson distribution, choose “Distribution”
from the “Analyze” menu. The variable we are wanting to look at is
“Number of events per week”, so put that under “Y, columns”. The total
number of data points that have that value of Y is listed in “Frequency - inhospital cardiac events”, so put that column into the “Frequency” box. The
box should look this, and then click “OK”.
In the window which opens in response, click on the red triangle next to
“Number of events per week” and choose “Discrete Fit” and then
“Poisson”:
32
Biology 300 lab manual, Lab 5
This will generate a Poisson distribution which has the same mean as the
data and plot it over the histogram. It will also open a new section at the
bottom of the window called “Fitted Poisson”. Click on the red arrow next
to “Fitted Poisson” and choose “Goodness of Fit”. This will tell JMP to run
a Goodness of Fit test to the Poisson distribution. In this case it tells us
that “Prob > X2” is 0.7379, meaning that the P-value is 0.7379 for this
test.
33
Biology 300 lab manual, Lab 5
Questions
1. Most hospitals now have signs posted everywhere banning cell phone use.
These bans originate from studies on earlier versions of cell phones. In one such
experiment, out of 510 tests with cell phones operating at near-maximum power,
six disrupted a piece of medical equipment enough to hinder interpretation of
data or cause equipment to malfunction. A more recent study found zero
instances of disruption of medical equipment out of 300 tests.
a. For the older data, calculate the estimated proportion of equipment
disruption. What is the 95% confidence interval for this proportion
calculated by JMP?
b. For the data on the later cell phones, use JMP to calculate the
estimate of the proportion and its 95% confidence interval.
[Note: JMP, like most computer packages, uses the Wald method to
calculate confidence intervals for proportions, which is not very
accurate for proportions close to zero or one.]
2. It is difficult to tell what other people are thinking, and it may even be
impossible to find out how they are thinking by asking them. A series of studies
shows that we do not always know ourselves how our thought processes are
carried out.
A classic experiment by Nisbett and Wilson (1978) addressed this issue in a
clever way. Participants were asked to decide which of four pairs of silk stockings
were better, but the four stockings that they were shown side-by-side were in fact
identical. Nearly all participants were able, however, to offer reasons that they
had chosen one pair over the other.
They were presented randomly with respect to which was in which position.
However, of the 52 subjects who selected a pair of stockings, 6 chose the pair on
the far left, 9 chose the pair in the left-middle, 16 chose the pair in the rightmiddle, and 21 chose the pair on the far right. None admitted later that the
position had any role in their selection.
Do a hypothesis test to ask, was the selection of the stockings independent of
position? If so, describe the pattern seen.
3. Are cardiac arrests (or "heart attacks") equally likely to occur throughout the
year? Or are some weeks more likely than others to produce heart attacks? One
way to look at this issue is to ask whether heart attacks occur according to a
model where they are independent of each other and equally likely at all times -the Poisson distribution. The data file "Cardiac events.csv" contains data from
one hospital over five years. It records both heart attacks that occurred to
individuals outside of the hospital and also heart attacks that occurred to patients
already admitted to the hospital.
34
Biology 300 lab manual, Lab 5
Does the frequency distribution of out-of-hospital cardiac events follow a Poisson
distribution?
4. Many people believe that the month that person is born in can predict
significant attributes of that person in later life. Such astrological beliefs have little
scientific support, but are there circumstances in which birth month can have a
strong effect on later life? One prediction is that elite athletes will
disproportionately be born in the months just after the age cutoff for separating
levels for young players. The prediction is that those athletes that are oldest
within an age group will do better by being relatively older, and therefore will gain
more confidence and attract more coaching attention than the relatively younger
players in their same groups. As a result, they may be more likely to dedicate
themselves to the sport and do well later. In the case of soccer, the cutoff for
different age groups is generally August.
a. The birth months of soccer
players in the Under-20's
World Tournament are
recorded in the data file
"Soccer births.csv." Plot
these data. Do you see a
pattern? (Photo from
VensPaperie on flickr.)
b. The numbers of people
born in Canada by month is
recorded in the file “NHL
births.csv” (don’t be
confused by the fact that
this file also has data on
birth months of hockey players). Compare the distribution of birth
months of the soccer players to what would be expected by
chance, assuming that the birth data for Canada is a good
approximation for the population from which soccer players are
drawn. Do they differ significantly? Describe the pattern.
5. Return to the page from the activities section where you circled the letter "t" in
the paragraph from Darwin's Origin of Species. (If you haven’t already done this,
please stop reading here now and go back to do that first.)
The point of this exercise is to collect data on whether our brains perceive words
merely as a collection of letters or if sometimes our brains process words as
entities. The logic of this test is that, if words are perceived as units, rather than
as collections of letters, then we should be more likely to do so for common
words. Hence we will look at the errors made in scanning this paragraph, and ask
whether we are more (or less) likely to miss finding a "t" when it is part of a
common word.
35
Biology 300 lab manual, Lab 5
Compare your results to the answer key that your TA will provide that marks all
the instances of the letter "t". Note that the answer key marks all "t"s in red, but it
also puts boxes around some words. The boxes are drawn around all instances
of the letter "t" occurring in common words. "Common" is defined here as among
the top-twenty words in terms of frequency of use in English; of these six contain
one or more "t"s: the, to, it, that, with, and at. In this passage there are 94 "t"s,
and 29 are in common words.
Count how many mistakes you made finding "t"s in common words and in less
common words. Use the data you collect yourself; also please report your results
to the TA so that they can be analyzed with the rest of the class.
a. If mistakes are equally likely for each common and less common word
(as defined above), what fraction of mistakes would be predicted to be
with common words, using this passage?
b. Use the appropriate test to compare your results to the null
expectation.
36
Biology 300 lab manual, Lab 6
6. Contingency analysis
Goals

Test for the independence of two categorical variables.
Quick summary from text (Chpt 9 from Whitlock and Schluter)

A contingency analysis allows a test of the independence of two or more
categorical variables.

The most common method of contingency analysis is based in the χ2
Goodness-of-Fit test. It makes the same assumptions as that test in terms
of the minimum numbers of expected values per group.

Fisher's Exact Test allows contingency analysis on two-by-two tables,
where each variable has two possible categories. This test makes no
assumptions about the minimum expected values.
Learning the Tools
1. χ 2 contingency analysis
A contingency analysis can be done from the "Fit Y by X" window. Choose
one of the categorical variables for "X, Factor" (preferably the one that you
view as the explanatory variable), and the other for "Y, Response." JMP
will then plot a mosaic plot, as you have seen before, and below the
mosaic plot are the results of a contingency analysis in the section marked
“Tests”. The row marked "Pearson" gives the standard χ2 contingency
analysis result. (The row marked "Likelihood Ratio" gives the results of a
G-test. See section 9.5 in Whitlock and Schluter.)
One thing that JMP does not do is very important. JMP does not check
that the assumptions of the χ2 test are met. Specifically, it doesn’t check
that the expected values are at least 5 for at least 80% of the cells. (Also it
is necessary that all cells have an expected value greater than one.) If
these assumptions are not met, ignore the χ2 and likelihood results, and
use Fisher's exact test, given at the bottom of the window.
To see the expected values, use the Fit Y by X window. Click on the red
triangle next to “Contingency Table” and choose the option “Expected.”
The expected values will appear at the bottom of each cell.
37
Biology 300 lab manual, Lab 6
2. Using "Frequency."
As for the goodness of fit test, in JMP one can use a faster data format
that summarizes the frequency of individuals that all share the same
characteristics. For example, using the "Titanic" data, we could do a
contingency analysis comparing the sex of passengers to whether they
survived the wreck. In the way the data is already formatted, each
individual has his or her own row. Alternatively, these data could be
represented in the following data table:
When using a data table like this, the column "Number in category" (or
whatever it is that you name the column with that the frequency data) must
be added to the "Frequency" box on the "Fit Y by X" screen. This
approach will give exactly the same answers as the individual-based
method. (Try it on these data to see.)
3. Calculating odds ratios
Odds ratios (and other methods of describing the relationship between two
categorical variables) can be calculated in JMP. After plotting the mosaic
plot with "Fit Y by X," choose "Odds Ratio" from the red triangle  menu
at the top left of the window. The odds ratio and its 95% confidence
interval will appear at the bottom of the window.
38
Biology 300 lab manual, Lab 6
Activities
We'll do a straightforward exercise to see how χ2 behaves when the null
hypothesis is true, and then later we'll see what χ2 looks like when the null
hypothesis is false.
To do so, let's start by using some dice to create a sample. The variables are
boring here, but the point is to see how the process works. We'll have two
variables: whether a die rolls a ONE, and the hand used to roll the dice. We know
pretty much already that there should be no association between which hand is
used to roll a die and the outcome. So if we test a null hypothesis that "hand" and
"one" are independent, we ought to find that there is no association. In other
words, the null hypothesis of independence is true.
1. Roll 30 dice (or one die 30 times) with each hand, and record the number
of "one"s for each hand. Create a JMP file to record these data with two
columns: Hand ("left" or "right") and One ("yes" or "no").
2. Do a χ 2 contingency analysis on the results.
Most samples ought to not reject the null hypothesis, but about one in
twenty samples will reject the null hypothesis. In all of these cases the null
hypothesis will be true, but by chance 5% of the samples will get a P-value
less than 5%.
3. Repeat steps 1 and 2 many, many times.
You can do this by hand, but we recommend using a java applet to
simulate the data collection to speed it up. Go to
http://onlinestatbook.com/stat_sim/contingency/index.html and, after it
loads, click "Begin." This will open a window that will simulate this kind of
data.
Let's consider "left-handed rolls" as "Condition 1" and "right-handed rolls
as "Condition 2." The column P(S) is where we enter the desired
probability for each case. Enter 0.1666 (i.e. one-sixth) in the box for P(s)
for both "conditions" in the top left. Enter 50 in both boxes for sample sizes
there as well. The window should look something like this, with four boxes
changed from their defaults:
39
Biology 300 lab manual, Lab 6
Now click the box called "Simulate 1." This will do the same kind of
sample that you just did with the dice, with the results appearing in the top
right corner. This applet will also calculate the χ2 for you.
Now click "Simulate 5000" one or more times. This will make 5000
separate samples, and calculate a χ2 for each. The program will keep
track of how many tests were statistically significant and how many were
not (see the bottom right sector of the window). You'll probably find that
about 5% of the results are "significant," meaning that the null hypothesis
would have been rejected. This is as expected; the significance level of
5% mean that we expect 5% of tests of true null hypotheses to reject the
null.
4. Change the true proportion of one of the P(s) in the simulation.
Using the applet still, let's simulate a loaded die, where the dice in the
rolled in the right hand are much more likely to roll a one. Let's change the
probability in the first row to be 0.35, but leave the second row set at
0.1666. Keep the sample size the same. With these settings, the null
hypothesis tested by the contingency analysis is now false: there is a
different probability of success in the two different conditions.
Before running the simulations, think for a moment about what you would
expect the results to look like. Should more or fewer of the samples give
high χ2 values and reject the null hypothesis? Will all samples reject the
null hypothesis, given that the null hypothesis is false?
Now run 5000 or so simulations. Compare what you see to what you
expected.
40
Biology 300 lab manual, Lab 6
Questions
1. Child seats have been a great success in reducing injuries and fatalities to
infants and toddlers under the age of two. This success has encouraged
governments to require child seats for older children as well.
The data file called "child seats-restraints only.csv" contains data about the
consequences of car accidents when children ages two to six are involved in an
accident severe enough that at least one other person is killed (Levitt 2005). This
data file includes data on kids who were in car seats, who were in lap and
shoulder belts, and who were in lap belts alone. The file excludes data on
children who were wearing no restraint; the data is clear that wearing no restraint
is substantially more dangerous that any of these other options.
The column called "Severity" indicates how severe the injuries are to the child.
The "Serious" category includes both death and incapacitating injuries.
a. Using these data, is there a significant difference in the severity of the
result between different methods of restraint?
b. Describe the results qualitatively. What policy decisions should be
made on the basis of these data?
c. Calculate the odds ratio for Severity, comparing child seats to lap belts
only.
2. Caribbean spiny lobsters normally prefer to live in shelters with other lobsters.
However, this may be a bad idea if the other lobster is infected with a contagious
disease. An experiment was done (Behringer et al. 2006) to test whether lobsters
can detect disease in other lobsters. Healthy lobsters were given a choice
between an empty shelter and a shelter containing another lobster. Half of the
experiments were done when the other lobster was healthy, and in the other half
the other lobster was infected with a
lethal virus (but before it showed
symptoms visible to humans).
Of 16 lobsters given a choice between
an empty shelter and one with a
healthy other lobster, seven chose the
empty shelter. However, of 16 given a
choice between the empty shelter and
sharing with a sick lobster, 11 chose
the empty shelter. Is there statistical
evidence that the lobsters are avoiding
the sick lobsters?
3. Return to the data you collected on
your class during the first week. Are dominant hand and dominant foot
correlated?
41
Biology 300 lab manual, Lab 6
4. Human names are often of obscure origin, but many have fairly obvious
sources. For example, "Johnson" means "son of John," "Smith" refers to an
occupation, and "Whitlock" means "white-haired" (from "white locks"). In
Lancashire, U.K., a fair number of people are named "Shufflebottom," a name
whose origins remain obscure.
Before children learn to walk, they move around in a variety of ways, with most
infants preferring a particular mode of transportation. Some crawl on hands and
knees, some belly crawl commando-style, and some shuffle around on their
bottoms.
A group of researchers decided to ask whether the name "Shufflebottom" might
be associated with a propensity to bottom-shuffle. To test this, the compared the
frequency of bottom-shufflers among infants with the last name "Shufflebottom"
to the frequency for infants named "Walker." (By the way, this study, like all
others in these labs, is real. See Fox et al. 2002.)
They found that 9 out of 41 Walkers moved by bottom-shuffling, while 9 out of 43
Shufflebottoms did. Is there a significant difference between the groups?
5. Falls are extremely dangerous for the elderly; in fact many deaths are
associated with such falls. Some preventative measures are possible, and it
would be very useful to have ways to predict which people are at greatest risks
for falls.
One doctor noticed that some patients stopped walking when they started to talk,
and she thought that the reason might be that it might be a challenge for these
people to do more than two things at once. She hypothesized that this might be a
cue for future risks, such as for falling, and this led to a study of 58 elderly
patients in a nursing home.
Of these 58 people, 12 stopped walking to talk, while the rest did not. Of the
people who stopped walking to talk, 10 had a fall in the next six months. Of the
other 46, 11 had a fall in that same time period.
a. Do an appropriate hypothesis test of the relationship between "stops
walking while talking" and falls.
b. What is the odds ratio of this relationship?
42
Biology 300 lab manual, Lab 7
7. Normal distribution and inference about means
Goals

Visualize properties of the normal distribution.

See the Central Limit Theorem in action.

Calculate sampling properties of means.
Quick summary from text (Chpts 10 + 11 from Whitlock and Schluter)

The normal distribution is the "bellshaped curve." It approximately
describes the distribution of many
biological variables.

The normal distribution has a single
mode equivalent to its mean, and it is
symmetric around the mean.

The normal distribution has two
parameters, the mean and the variance.

If a distribution is approximately normal (bell-shaped), then about twothirds of the data will fall within one standard deviation above or below the
mean. About 95% will all within two standard deviations above or below
the mean.

The Central Limit Theorem states that the sum or mean of a set of
independent and equally distributed values will have a normal distribution,
as long as enough values are added together. This means that the mean
of large sample will be normally distributed, even if the distribution of the
variable itself is not normal.

The confidence interval for a mean assumes that the variable has a
normal distribution in the population and that the sample is a random
sample.

The 95% confidence interval for the population mean is approximately 2
standard errors above and below the sample mean.

The one-sample t-test is a hypothesis test that compares the mean of a
population to a hypothesized constant. The one-sample t-test makes the
same assumptions as the confidence interval.
43
Biology 300 lab manual, Lab 7
Activity 1: Give the TA the finger(s)
Using "Student Data Sheet 7" at the end of this manual, please make the
suggested measurements and record them on the sheet. Then, if you don't mind
sharing these data with the class, give the sheet to your TA who will compile
them for use in Question 2.
Activity 2: Distribution of sample means
We return to the applet we used in chapter 3, located at
http://onlinestatbook.com/stat_sim/sampling_dist/index.html. We want to
investigate three claims made in the text.
Claim 1: The distribution of sample means is normal, if the variable itself
has a normal distribution.
First, hit "Animated" a few times, to remind yourself of what this applet
does. (It makes a sample from the distribution shown in the top panel. The
second panel shows a histogram of that sample, and the third panel
shows the distribution of sample means from all the previous samples.)
Next, hit the "10,000" button. This button makes 10,000 separate samples
at one go, to save you from making the samples one by one.
Look at the distribution of sample means. Does it seem to have a normal
distribution? Click the checkbox by "Fit normal" off to the right, which will
draw the curve for a normal distribution with the same mean and variance
as this distribution of sample means.
Claim 2: The standard deviation of the distribution of sample means is
predicted by the standard deviation of the variable, divided by the square
root of the sample size.
The standard deviation of the population distribution in the top panel is
given in the top left corner of the window, along with some other
parameters. The sample size is set by the pull-down menu on the right of
the sample mean distribution. (The default when it opens is set to N=5.)
For N=5, have the applet calculate 10,000 sample means as you did in the
previous exercise. If the sample size is 5 and the standard deviation is 5.0
(as in the default), what do you predict the standard deviation of the
sample means to be? How does this match the simulated value?
Change the sample size to N=25, and recalculate 10,000 samples.
Calculate the predicted standard deviation of sample means, and compare
it to what you observed.
44
Biology 300 lab manual, Lab 7
Claim 3: The distribution of sample means is approximately normal no
matter what the distribution of the variable, as long as the sample size is
large enough. (The Central Limit Theorem).
Let's change the distribution of the variable in the population. At the top
right of the window, change "Normal" to "Skewed." This will cause the
program to sample from a very skewed distribution.
Set N=2 for the sample size, and simulate 10,000 samples. Does the
distribution of sample means look normal? Is it closer to normal than the
distribution of individuals in the population?
Now set N=25 and simulate 10,000 samples. How does the distribution of
sample means look now? It should look much more like a normal
distribution, because of the Central Limit Theorem. (Explain in your own
words why this is so.)
Finally, look at the standard deviations of the distribution of sample means
for these last few cases, and compare them to the expectation from the
standard deviation of individuals and the sample size.
If you want to play around with this applet, it will let you draw in your own
distribution at the top. Just hold down the mouse button over the top
distribution, and it will let you paint in a new distribution to use. Try to
make a distribution as non-normal as possible, and then draw samples
from it.
Learning the Tools
1. One-sample t-tests
For an example, let's use the data in "titanic.csv" on the passengers of the
Titanic. We'll ask whether the mean age of passengers was significantly
different from 18 years old. The first step is to use the "Distribution"
window to create a histogram and summary statistics. (You find
"Distribution" under the "Analyze" menu bar.) Choose "age" as the
variable, and hit OK.
To do a one-sample t-test, click on the red triangle  next to the variable
name, "age." From the menu that appears, choose "Test Mean." A window
will open that asks for the mean proposed by the null hypothesis. In our
case, that value should be 18. Enter the value and click OK.
45
Biology 300 lab manual, Lab 7
This will open a new section in the results window, which will give the
results of a one-sample t-test, with the value of the test statistic t, the
degrees of freedom, and the P-value for one- and two-sided tests.
2. Checking for normality
One of the best ways to know whether a population is approximately
normal is to plot a histogram of the data. If the sample size is large and
the data look even approximately normally distributed, then that will be
close enough for most purposes. JMP will let you draw a normal
distribution with the same mean and standard deviation as the data on top
of the histogram, to make it easier to visualize. In the menu under the red
triangle  by the variable name, click "Continuous Fit" and choose
"Normal" from that menu.
To more formally look at normality, make a quantile plot. After you used
"Continuous Fit," a new section called "Fitted Normal" will have appeared
at the bottom of the window, with a new red triangle . Under that
triangle, click "Diagnostic Plot." If a data set is approximately normal, then
the points in the quantile plot will mainly fall along a straight line.
46
Biology 300 lab manual, Lab 7
These age data from the Titanic passengers are significantly different from
a normal distribution. However, the shape of the distribution is not that
different from a normal distribution, as you can see because the points
mainly fall along the diagonal line in this quantile plot. Therefore, most
methods that assume normality would work well in this case.
3. Transformations
In some cases a distribution will be more normal if the data are
transformed before analysis, for example by taking the log of each data
point. To do so, make a new column in the data file, and in this column
use Formula to create a transformation. For example, to do a logtransformation, in the Formula window choose "Log" under the
Transcendental menu (under Functions), and then click on the variable
you want to log to put it into the formula. For example, to log-transform
age, here is what the formula will look like:
After the formula is entered, hit "Apply" and "OK". You can then analyze
the new variable just like any other variable in JMP.
47
Biology 300 lab manual, Lab 7
Questions
1. Insects like the fruit fly Drosophila
melanogaster have immune systems, but they
rely on innate immunity rather than acquired
resistance to specific antigens. As such, the
genetic variation in immune response is
extremely important for understanding the effects
of disease on fly populations. A large number of
flies were measured for their ability to suppress
growth of a gram-negative pathogenic bacteria.
Each fly was inoculated with the disease, and
then the amount of bacteria in each was
measured later. The log-bacteria per fly is listed
under the column “Bacterial load” in the file “antibacterial immunity.csv”.
a. Use at least two methods to determine whether bacterial load is
approximately normally distributed.
b. On average, is the mean of the bacterial load score different from
zero? Do the appropriate statistical test.
c. Calculate a 95% confidence interval for the mean bacterial load score
of these flies.
d. Use the mean +/- 2 standard error approximation to calculate a 95%
confidence interval for the mean score. Compare the result of this
approximation to the more precise answer from part (b).
2. Earlier this lab you will have measured your fingers. We're going to use this
data now and also in the next lab. We'll look at the ratio of the lengths of index
finger over ring finger. This is called the 2D:4D ratio, where the 2D refers to
second digit and 4D means fourth digit.
It turns out—bizarrely—that this 2D:4D ratio is a measure of
the amount of testosterone that a fetus is exposed to during
gestation. Next week we'll test whether that corresponds to
differences for males and females. For this week, we'll estimate
the 2D:4D ratio, and ask whether its mean is significantly
different from 1. If index fingers are on average equal to ring
fingers in length, then the ratio would be one.
First, use the finger data file your TA has compiled. Add two
new columns, and in these columns use "Formula" to create
the new variables "Right 2D:4D ratio" and "Left 2D:4D ratio."
a. What is the 95% confidence interval for "Right 2D:4D ratio"? For the
left ratio?
48
Biology 300 lab manual, Lab 7
b. For each of the two ratios, test whether the mean ratio differs from one.
c. Compare the results from the confidence interval and the hypothesis
test. Are they consistent?
3. Mosquitoes find their victims in part by odor, so it makes sense to wonder
whether what we eat and drink influences our attractiveness to mosquitoes. A
study in West Africa (Lefèvre et al. 2010), working with the mosquito species that
carry malaria, wondered whether drinking the local beer influenced attractiveness
to mosquitoes. They opened a container holding 50 mosquitoes next to each of
25 alcohol-free subjects and measured the proportion of mosquitoes that left the
container and flew toward the subject (they called this proportion the “activation”).
They repeated this procedure 15 minutes after each of the same subjects had
had a liter of beer, and measured the “change in activation” (after minus before).
The data in the file “BeerAndMosquitos.csv” are the values of mosquito activation
(fraction of 50 mosquitos that left the container) recorded on the 25 subjects
before and after drinking 1 liter of the local beer.
a. Calculate a 95% confidence interval for the difference in activation
after drinking beer.
b. Does having beer change the average attractiveness of a person to
mosquitoes? Do an appropriate hypothesis test.
4. The file "mammals.csv" contains information on the body size and brain size of
various mammal species.
a. Plot the distribution of brain size, and describe its shape. Does this
look like it has a normal distribution?
b. Use a histogram to examine whether brain size has a normal
distribution.
c. Transform the brain size data with a log-transformation. Plot the
distribution of log brain size. Describe the new distribution, and
examine it for normality.
Assignment (this may change—see instructors for exact
assignment)
Using the data collected by all individuals in your class in response to the Activity
in Lab 5, test whether there is evidence that people in general are likely to read
common words as blocks rather than as a collection of letters (in other words, are
"t"s in common words found differently than those in uncommon words). Write up
the results in a journal article format, with an introduction, methods and materials,
results, and discussion. Include both graphical and statistical representations of
the data. The report should be approximately 3-4 double-spaced pages. The TA
will give you a copy of the data for the whole class to work from.
49
Biology 300 lab manual, Lab 8
8. Comparing two groups
Goals

Compare the means of two groups

Understand paired vs. two-sample designs

Explore the effects of violation of assumptions
Quick summary from text (Chpt. 12 from Whitlock and Schluter)

The most common method to compare the means of two groups is the
two-sample t-test. The two-sample t-test assumes that there is an
independent random sample from each group, and that the variable is
normally distributed in each population with the same variance.

If each data point in one group is uniquely paired with a data point in the
other group, then the data are said to be paired. A paired t-test examines
the mean difference between the members of each pair. Paired t-tests
assume that the differences are normally distributed and that the pairs are
randomly sampled.

Welch's t-test is similar to a two-sample t-test, but does not require the
assumption of equal variances.

Most hypothesis tests make assumptions about the distribution of the
population that the data are taken from, for example, that the variable has
a normal distribution. Such tests are called parametric tests.

Many parametric tests, including the t-test, are robust to their assumptions
This means that the test works well even when the assumptions are
violated. For example, with large sample sizes, the 2-sample t-test works
acceptably even if the two groups’ variances differ by a factor of 10.

Data can sometimes be transformed to make new variables that better
match the assumptions of the methods. For example, with the logtransformation, the natural logarithm of the value of a variable is used
instead of the variable itself.
50
Biology 300 lab manual, Lab 8
Activities
1. Collect some data on your dexterity
Using "Student Data Sheet 8", at the end of this manual, collect the data
on the speed of each hand, and record them on the sheet. Then, if you
don't mind sharing these data with the class, give the sheet to your TA
who will compile them for use in Question 3.
2. Investigating robustness
The t-test is fairly robust to its assumptions. This means that even when
the assumptions are not correct, the t-test often performs quite well.
Remember that the two-sample t-test assumes that the variables have a
normal distribution in the populations, that the variance of those
distributions is the same in the two populations, and that the two samples
are random samples.
Open a browser and load the applet by clicking "Begin" at
http://onlinestatbook.com/stat_sim/robustness/index.html. This applet will
simulate t-tests from random samples from two populations, and it lets you
determine the true nature of those populations.
Start with a scenario that matches the assumptions of the t-test. Use the
default parameters of Mean = 0, standard deviation (sd) = 1, and no skew
for both populations. (These values can be changed by the controls on the
left hand side of the window.) You can leave the sample sizes at 5
(controls for the sample sizes are in the right side of the window.) Now
click "Simulate" at the bottom. The computer will artificially create 2000
samples of 5 individuals each from both populations, and calculate the
results of a two-sample t-test comparing the means. It tallies the numbers
of significant and non-significant results at the bottom of the window. In
this case, both distributions are normal and the variances are equal, just
as assumed by the t-test. Therefore we expect it to work quite well in this
case.
(Take a moment to think about what "working well" means. If the null
hypothesis is true, then the Type I error rate should match the stated Type
I error rate, and if the null hypothesis is false it should be rejected as often
as possible.)
Now, look at the following cases, and describe the effects on the
performance of the test.
a. (Unequal standard deviation, equal sample size) Make the standard
deviations of the two populations unequal, with the first equal to one
and the second equal to 5, with everything else like the defaults. (In
particular, keep the sample sizes equal.)
51
Biology 300 lab manual, Lab 8
b. (Unequal standard deviation, unequal sample size) Make the
standard deviations unequal as in part a. (that is, 1 and 5), but make
the sample size of the first population be 25, leaving the second
sample with 5 individuals.
c. (Skew, equal sample size) Return to the defaults, except make both
populations have "Severe" skew.
Play around with other parameter combinations. What can you conclude
about the robustness of the two-sample t-test?
Learning the Tools
1. Paired t-test
To do a paired t-test in JMP, the data have to be stored in a particular
way. Each pair needs to have a single row, with the value of one member
in the pair in one column and the value for the other member in another
column. For example, if the paired t-test was comparing a variable before
and after a treatment, there would be one column for the variable before
treatment and another column for after.
There are two different ways to do a paired t-test in JMP. The first option
is to create a new column and use "Formula" to calculate the difference
between the two columns for the variable, for example after minus before.
The difference column can then be analyzed with a one-sample t-test
using the "Distribution" window." (This works because a paired t-test is just
a one-sample t-test on the differences.)
JMP will also do a paired t-test directly, using the "Matched Pairs" window,
available under the "Analyze" menu bar. Choose the two variables that
represent the two members of each pair (for example "before" and "after")
for the "Y, paired response" blank, and hit OK.
The resulting graph is rather confusing, but in essence it plots the
difference in values against the mean of the two members of each pair.
The results of a paired t-test are given at the bottom of this window.
For example, the data file "NZNoggins.csv" contains information on head
sizes of New Zealand army recruits, measured either with metal calipers
or a cheaper cardboard method. Each row represents a different soldier,
measured in each of the two ways. The results show that the cardboard
method is biased, giving on average larger values than the metal calipers.
52
Biology 300 lab manual, Lab 8
2. Making a dot plot
Making a dot plot in JMP takes a little work. Let’s use the titanic data set,
to make a dot plot of the age of those who survived and those who died.
With the titanic data set open, choose “Fit Y by X” from the “Analyze”
menu. Choose “Survived” as the X variable and “Age” as the Y variable,
and hit OK. In the resulting window with the graph, choose “Means and
Std Dev” from the menu at the red triangle, which will add the means and
confidence bars on the graphs. Next, choose “Display options” from the
red triangle menu, and select “Points jittered”. This will spread the points
out a bit so that you can see more about where the data are.
The result is not the best dot plot you have ever seen, but it gives a sense
of the data.
3. Two-sample t-test
To do a two-sample t-test, first plot the numerical variable for both groups
using “Distribution” under the “Analyze” menu. Choose your numerical
variable as Y, and put your group variable in the box by “ By”. This will plot
a histogram separately for both groups.
You can analyze the data with "Fit Y by X" from the "Analyze" menu. For
example, let's compare passengers on the Titanic that survived to those
that died, to ask whether they differ in age. (We might expect that they do,
because of the social norm to put "women and children first.") Choose
"Survived" as the X variable, and "age" as the Y variable in the "Fit Y by X"
screen.
Next, from the red triangle menu at the top of the window, choose
"Means/ANOVA/Pooled t." [Important note: Confusingly, don’t choose "t
Test," which will do a Welch's t-test, as we discuss in the next section.]
53
Biology 300 lab manual, Lab 8
The results section that this will create contains a lot of information. Here
are the highlights:
4. Welch's t-test (not assuming equal variances)
Welch's t-test does not assume that the variance is the same in the two
groups. To do a Welch's t-test in JMP, first open the "Fit Y by X" window,
but then choose "t Test" under the red triangle. The results section that
opens looks like the following, including the phrase "assuming unequal
variances."
Alternatively, under the red triangle menu, choose "Unequal variances."
This does a test for the equality of variances and performs Welch's t-test
automatically, as shown in the following result screen.
54
Biology 300 lab manual, Lab 8
Questions
1. A common belief is that shaving causes hair to grow back faster or coarser
than it was before. Is this true? Lynfield and McWilliams (1970) did a small
experiment to test for an effect of shaving. Five men shaved one leg weekly for
several months while leaving the other leg unshaved. The researchers measured
the change in leg hair width and the hair growth rate on each of the legs. These
data are given in "leg shaving.csv."
a. Find the 95% confidence interval for the difference between shaved
and unshaved legs for both variables.
b. Perform a suitable hypothesis test to ask whether shaving affects hair
width.
c. Perform a suitable hypothesis test to ask whether shaving affects hair
growth rate.
2. In the last lab, we collected data on the lengths of your second and fourth
fingers (2D and 4D -- your TA will tell you where to find the data.) As mentioned
last time, the 2D:4D ratio is a measure of the amount of testosterone that a
human is exposed to as a fetus during the period of finger development. As such,
we might expect there to be a difference between males and females in the
2D:4D ratio. Use these data to calculate the 2D:4D ratio (index finger length
divided by ring finger length) on the right hands.
a. Test for a difference in the mean 2D:4D ratio between men and
women.
b. What is the 95% confidence interval for the difference in means?
3. Return to the data collected early in the lab on dexterity of each hand, which
should be collated by the TA. Each of you has measured your dexterity with each
hand. Do a hypothesis test to compare average difference in the dexterity of the
dominant hand compared to the less dominant hand.
55
Biology 300 lab manual, Lab 8
4. London taxi drivers are required to learn 25,000 streets within a 6-mile radius
of central London -- a massive compendium of spatial information referred to as
The Knowledge. Spatial information is thought to be processed in a part of the
brain known as the posterior hippocampus. (This part of the brain is larger in bird
and mammal species that use a great deal of spatial information, such as
species that store caches of food.)
Maguire et al. (2000) wanted to know whether this
great skill in spatial navigation translated into
changes in brain morphology in London cabbies.
They used MRI to measure the cross-sectional area
of the posterior hippocampus in two groups of cab
drivers: one group who had been driving for less
than 15 years, and another group who had been
driving for more than 15 years. The data are in the
file "hippocampus.csv."
a. Test for a difference in the mean posterior hippocampus size between
newer and older drivers.
b. What is the 95% confidence interval for the difference in means?
5. In 1898, Hermon Bumpus collected data on one of the first examples of natural
selection directly observed in nature. Immediately following a bad winter storm,
136 English house sparrows were collected and brought indoors. Of these, 72
subsequently recovered, but 64 died. Bumpus made several measurements on
all of the birds, and he was able to demonstrate strong natural selection on some
of the traits as a result of this storm.
Bumpus published all of his data, and they are given in the file "Bumpus.csv."
Test whether the birds that survived differ from the dead birds for each of the
following traits:
a. Total length
b. Sex
c. Skull width
d. Beak and Head length
e. Weight
f. Age
56
Biology 300 lab manual, Lab 9
9. Comparing more than two groups: ANOVA
Goals

Compare the means of multiple groups using Analysis of Variance.

Make post-hoc comparisons with the Tukey-Kramer test.
Quick summary from text (Chpt. 15 in Whitlock and Schluter)

Analysis of variance (ANOVA) compares the means of multiple groups.

A significant ANOVA result implies that at least one group has a different
mean from one other group.

Tukey-Kramer tests allow comparison of each pair of means.

ANOVA assumes equal variance of each group, each group has a normal
distribution of the variable, and each population is randomly sampled.

The Kruskal-Wallis test is a nonparametric test that compares multiple
groups, often used in place of ANOVA when ANOVA's assumptions aren't
true.
Learning the Tools
1. ANOVA
To do a one-way ANOVA, we can use the "Fit Y by X" routine from the
"Analyze" menu bar. The categorical or group variable should be entered
as "X, factors" and the response variable should be entered as "Y,
response." When the results window opens with a dot plot of the data,
click on "Means/Anova" under the red triangle menu in the top left corner.
For example, let's use the "15e1KneesWhoSayNight.csv" data set that
correspond to example 15.1 in Whitlock and Schluter. "Treatment" is the
group variable to put in the "X" box and "shift" is the numerical response
variable to put in as "Y." Doing the ANOVA gives the following window of
results.
57
Biology 300 lab manual, Lab 9
2. Tukey-Kramer tests
JMP 7: Tukey-Kramer tests can also be done from the menu under the red
triangle. Click on "Compare means" and then on "All Pairs, Tukey HSD."
This will open a results section that looks something like this:
58
Biology 300 lab manual, Lab 9
3. Kruskal-Wallis test
A Kruskall-Wallis test can be performed under an alternative name, the
Wilcoxon test. In the menu under the red triangle at the top left of the
window, click on "Nonparametric" and then choose "Wilcoxon Test." A
results section at the bottom of the window will show the results. The Pvalue of the test is given under the heading of "Prob >ChiSq."
Questions
1. The pollen of the corn (maize) plant is known to be a source of food to larval
mosquitoes of the species Anopheles arabiensis, the main vector of malaria in
Ethiopia. The production of maize has increased substantially in certain areas of
Ethiopia recently, and over the same time malaria has entered in to new areas
where it was previously rare. This raises the question, is the increase of maize
cultivation partly responsible for the increase in malaria?
One line of evidence is to look for a geographical association on a smaller spatial
scale between maize production and malaria incidence. The data set "malaria vs
maize.csv" contains information on several high-altitude sites in Ethiopia, with
information about the level of cultivation of maize (low, medium or high) and the
rate of malaria per 10,000 people.
a. Plot the relationship between maize production and the incidence of
malaria.
b. Do these data seem to match the assumptions of ANOVA? If not, do
any simple transformations solve the problem?
c. Test for an association between maize yield and malaria incidence.
2. The European cuckoo does not look after its own eggs, but instead lays them
in the nests of birds of other species. We have seen previously that cuckoos
sometimes have evolved to lay eggs that are colored similarly to the host birds'
eggs. Is the same true of size -- do cuckoos lay eggs of different sizes in nests of
different hosts? The data file "cuckooeggs.csv" contains data on the lengths of
cuckoo eggs laid in a variety of other species' nests.
a. Test for a difference between the host species in the size of the cuckoo
eggs.
b. Using a Tukey-Kramer test, which pairs of host species are
significantly different from each other?
59
Biology 300 lab manual, Lab 9
3. Animals that are infected with a pathogen often have disturbed circadian
rhythms. (A circadian rhythm is a daily cycle in a behavior or physiological trait
that persists in the absence of time cues.) Shirasu-Hiza et al. (2007) wanted to
know whether it was possible that the circadian timing mechanism itself could
have an effect on disease. They used three groups of fruit flies: one "normal",
one with a mutation in the timing gene tim01, and one group that had the tim01
mutant in a heterozygous state. They exposed these flies to a dangerous
bacteria, Streptococcus pneumoniae, and measured how long the flies lived
afterwards, in days. The date file "circadian mutant health.csv" shows some of
their data.
a. Plot a histogram of each of the three groups. (Hint: you can use the
"By" function in the Distribution window to specify the group.) Do these
data match the assumptions of an ANOVA?
b. Do an appropriate hypothesis test to ask whether lifespan differs
between the three groups of flies.
60
Biology 300 lab manual, Lab 10
10. Regression and correlation
Goals
Use the computer to calculate regression lines and correlation coefficients, test
null hypotheses about slopes, and deal with non-linear data.
Quick summary (Chpts 16 + 17 from Whitlock and Schluter)

Linear regression uses a line to predict a response numerical variable (Y)
from a numerical explanatory variable (X).

The regression line will take the form Y = a + b X. a is the intercept (where
the line crosses the axis at X = 0), and b is the slope of this regression
line.

Linear regression assumes that the true relationship between X and Y
follows a line, that the data are taken from a random sample, and that for
every value of X the corresponding values of Y are normally distributed
with the same variance for all X.

These assumptions can be examined with a residual plot, where the
residual (deviation between the Y value and its predicted value) is plotted
against X.

Hypothesis tests about regression slopes can be made using either a t
test or an ANOVA-like approach.

Confidence bands show the limits of confidence intervals for the mean Y
for each value of X. Prediction intervals show confidence intervals for the
individual values of Y for each X.

R2 measures the fraction of variance in Y that is predicted by X.

Non-linear relationships between X and Y can often be examined by use
of transformations.

The correlation coefficient (r) measures the strength of association
between two numerical variables, X and Y.

A positive r means that larger values of X are associated with, on average,
larger values of Y. A negative r means that large X values and small Y
values tend to appear together.

Hypothesis tests and confidence intervals for the correlation coefficient
assume that X and Y have a linear relationship and that the sample used
61
Biology 300 lab manual, Lab 10
is a random sample. These methods also assume that for any given value
of X, Y is normally distributed with equal variance. Also, for any value of Y,
X is assumed to be normally distributed with equal variance.
Learning the Tools
1. Linear regression
We'll illustrate the steps of using the program to do regression with
Example 17.1, predicting the age of a lion from the amount of black on its
nose pad. In the data file, the age of lion is called age, given in years, and
the proportion of black on its nose pad is called proportion.black.
Load the data from the file "17e1LionAges.csv". Make a scatter plot of
the relationship between age and proportion.black, using "Fit Y by X" from
the Analyze menu bar. Have age be the response variable and
proportion.black be the explanatory variable.
To calculate the regression line and test the null
hypothesis that the slope is zero, click on the small red
triangle at the top left of the scatter plot window. From the
menu that opens, choose "Fit Line." This will increase the
size of the window, with new results added below the
scatter plot. These results include the equation for the
best-fitting line, the R2 value for the regression, and
hypothesis tests of the null hypothesis that the slope of the line is zero.
The hypothesis test is done both by the ANOVA approach and by using
the t-test.
62
Biology 300 lab manual, Lab 10
2. Residual plots
Next, examine the residuals plot for deviations from linearity and other
assumptions. Click on the red triangle  next to "Linear Fit," right under
the graph. From the menu that appears, choose "Plot Residuals." A
residual plot will appear at the bottom of the window. Look for evidence of
unequal variance around the line or patterns that may indicate a non-linear
relationship. For the lion data, there is no strong reason to doubt the
assumptions of the analysis are OK.
63
Biology 300 lab manual, Lab 10
3. Transform data as required.
This example does not require transformation. If data does need a
transformation, you can easily do it using JMP. First, create a new column
in the spreadsheet, and then use the "Formula" function to calculate
transformed values in the new column. See chapter 7 for discussion of
transformations in this way.
4. Calculate confidence bands or prediction intervals.
Confidence bands for the mean predicted Y can be printed on the graph
by choosing "Confid Curves Fit" from the same menu under the red
triangle  at "Linear Fit." Prediction intervals (showing the confidence
intervals for predictions of individual values of Y) can fit by clicking "Confid
Curves Indiv" from the same menu. The curves closest to the lines will be
the confidence bands for the mean Y; those further from the line will
indicate the individual prediction intervals.
64
Biology 300 lab manual, Lab 10
5. Calculating correlation coefficients
The Pearson's correlation coefficient between two numerical variables can
be calculated using the "Multivariate Methods  Multivariate" menu
choice under the "Analyze" menu. Choose the variables that you want to
correlate by putting them in the "Y, columns" list, and hit OK.
To make statistical tests about correlation coefficients, click on the red
triangle by the "Multivariate" title, and choose "Pairwise Correlations." A
results section will appear with the correlation between each pair of
variables and the P-value of a test of the null hypothesis that they are not
correlated.
In this display, “Correlation” is the correlation coefficient r, and “Signif
Prob" is the P-value for the test that this correlation coefficient is zero.
Activities (Optional)
Use the applet at http://onlinestatbook.com/stat_sim/transformations/index.html
to explore the effects of transformations on the various data sets provided there.
Click the transformation buttons on the side and bottom of each graph, and
select alternative data sets from the menu button at the top.
Questions
1. The ends of chromosomes are called telomeres. These telomeres are
shortened a bit during each cell cycle as DNA is replicated; one of their purposes
is to protect more valuable DNA in the chromosome from degradation during
replication. As individuals age, their telomeres shorten, and there is evidence that
shortened telomeres may play a role in aging. Telomeres can be lengthened in
65
Biology 300 lab manual, Lab 10
germ cells and stem cells by an enzyme called telomerase, but this is not
expressed in most healthy somatic cells. (Cancer cells, on the other hand,
usually express telomerase.)
Given that the length of telomeres is biologically important, it becomes interesting
to know whether telomere length is inherited. A set of data was collected by
Nordfjäll et al. (2005) on the telomere length of fathers and their children; these
data are in the file "telomere inheritance.csv."
a. Plot a scatter plot showing the relationship between father and
offspring telomere length.
b. Do the data require any transformation before analysis?
c. Create an equation that predicts the offspring telomere length from its
father's. Is there evidence that the father's telomere length predicts his
offspring's value?
d. How many offspring values lie outside the prediction intervals for this
regression line? Is this more than you would expect by chance?
Explain.
2. Opfer and Segler (2007) asked several school children to mark on a number
line where a given number would fall. The child was given a number line with two
ends marked at 0 and 1000, and then asked to make an X on that line where a
number, for example 150, should be placed. They asked the children to place
several different numbers on the number lines, each on a fresh new piece of
paper.
Two separate groups of kids were asked to do this task: one group of secondgraders and another group of fourth-graders. The researchers then measured
each mark to see the value on a linear scale of its placement. The results,
averaged over all 93 kids for each group, are given in the file "numberline.csv".
a. Plot the fourth graders' guesses against the true value. Is this
relationship linear? If not, find a transformation of X or Y that describes
the data well.
b. Plot the second-graders' guesses against the true value. Is this
relationship linear? If not, find a transformation of X or Y that describes
the data well. Examine the residual plots for both the linear and nonlinear fits.
66
Biology 300 lab manual, Lab 10
c. Assume that the difference between the shapes of these curves is real.
What would you conclude about the difference in the way 2nd graders
and 4th graders perceive numbers?
3. Larger animals tend to have larger brains. But is the increase in brain size
proportional to the increase in body size? A set of data on mammal body and
brain size was collated by Allison and Cicchetti (1976), and theses data are to be
found in the data set "mammals.csv." The file contains columns giving the
species name, the body mass (in kg) and brain size (in g) of 62 different mammal
species.
a. Plot brain size against body size. Is the relationship linear?
b. Find a transformation (for either or both variables) that makes the
relationship between these two variables linear.
c. Is there statistical evidence that brain size is correlated with body size?
d. What line best predicts (transformed) brain size from (transformed)
body size.
e. If a mammal species had a log- body size that was three units greater,
how much would log-brain size be expected to change?
f. Make a residual plot. Which species has the highest brain size for their
body size? Which species has the smallest brain relative to that
predicted by its body size?
g. Are humans significantly different from the brain size predicted by our
body size?
h. Are chimpanzees significantly different from the brain size predicted by
their body size?
67
Biology 300 lab manual
References and Photo Credits
Chapter 1
countries data from : http://www.worldbank.org/
Muchhala, N. 2006. Nectar bat stows huge tongue in its rib cage. Nature 444:701-702.
Chapter 3
McCusker, R. R., B. A. Goldberger, and E. J. Cone. 2003. Caffeine content of specialty coffees.
Journal of Analytical Toxicology 27:520-522.
Chapter 5
soccer birth data: Barnsley, R. H., A. H. Thompson, and P. Legault. 1992. Family planning:
football style. The relative age effect in football. Int. Rev. for Soc. of Sport 27:77-86.
NHL birth data: http://www.cd-b.com/nhl2-1.htm
Birth months: Statistics Canada
Cardiac arrest distribution: Skogvoll, E., and B. H. Lindqvist. 1999. Modeling the occurrence of
cardiac arrest as a Poisson process. Ann. Emerg. Med. 33:409-17.
Position effects: Nisbett, R. E., and T. D. Wilson. 1977. Telling more than we can know: verbal
reports on mental processes. Psychological Review 84:231-259.
Nisbett, R. E., and T. D. Wilson. 1978. The accuracy of verbal reports about the effects of stimuli
on evaluations and behavior. Social Psychology 41: 118-131.
Chapter 6
Levitt, S. D. 2005. Evidence that seat belts are as effective as child safety seats in preventing
death for children aged two and up. Review of Economics and Statistics 90: 158-163.
Behringer, D. C., M. J. Butler, and J. D. Shields. 2006. Avoidance of disease by social lobsters.
Nature 441:421.
Caribbean spiny lobster photo: http://www.flickr.com/photos/sniffette/1677217581/
Mohammed, M. A., J. Mant, L. Bentham, A. Stevens, and S. Hussain. 2006. Process of care and
mortality of stroke patients with and without a do not resuscitate order in the West
Midlands, UK. International Journal for Quality in Health Care 18:102–106.
Fox, A. T., R. D. Palmer, and P. Davies. 2002. Do "Shufflebottoms" bottom shuffle? Arch. Dis.
Child. 87:552-554.
Lundin-Olsson, L., L. Nyberg, and Y. Gustafson. 1997. 'Stops walking when talking' as a predictor
of falls in elderly people. Lancet 349:617.
Chapter 7
Lazzaro, B. P., B. K. Sceurman, and A. G. Clark. 2004. Genetic basis of natural variation in D.
Melanogaster antibacterial immunity. Science 303:1873-1876.
Chapter 8
Lynfield, Y. L., and P. MacWilliams 1970. Shaving and hair growth. Journal of Investigative
Dermatology 55:170-172.
Maguire, E. A., et al. 2000. Navigation-related structural change in the hippocampi of taxi drivers.
PNAS 97: 4398 – 4403.
68
Biology 300 lab manual
Bumpus, Hermon C. 1898. Eleventh lecture. The elimination of the unfit as illustrated by the
introduced sparrow, Passer domesticus. (A fourth contribution to the study of variation.)
Biol. Lectures: Woods Hole Marine Biological Laboratory, 209-225.
Chapter 9
Latter, O.H. 1902. The eggs of Cuculus canorus. An Inquiry into the dimensions of the cuckoo's
egg and the relation of the variations to the size of the eggs of the foster-parent, with
notes on coloration, &c. Biometrika i, 164.
Kebede, A., J. C. Mccann, A. E. Kiszewski, and Y. Ye-Ebiyoam. 2005. New evidence of the
effects of agro-ecologic change on malaria transmission. J. Trop. Med. Hyg. 73:676–680.
Shirasu-Hiza, M., M. S. Dionne, L. N. Pham, J. S. Ayres, and D. S. Schneider. 2007. Interactions
between circadian rhythm and immunity in Drosophila melanogaster. Current Biology 17:
R353-R355.
Chapter 10
Allison, T., and D. Cicchetti. 1976. Sleep in mammals: Ecological and constitutional correlates.
Science 194:732-734.
Nordfjäll, K., Å. Larefalk, P. Lindgren, D. Holmberg, and G. Roos. 2005. Telomere length and
heredity: indications of paternal inheritance. PNAS 102:16374-16378.
Opfer, J., and R. Siegler. 2007. Representational change and children's numerical estimation.
Cognitive Psychology 55:169-195.
brain/body size data available at http://lib.stat.cmu.edu/datasets/sleep.
69
Biology 300 lab manual
Stats lab student data sheet 1
Record the following measurements on yourself. These will be shared (without
your name attached) with the rest of the class, so if you feel any discomfort with
such sharing feel free to omit any or all of the measurements.
Height (cm)
Circumference Measure around the
of head (cm)
head at eye-brow
level.
Sex
Number of
siblings
How many siblings
(not counting
yourself) are in your
family?
Handedness
Which hand do you
normally write with?
Dominant foot
Which foot do you
kick a ball with?
Dominant eye
Make a triangle with your index fingers and thumbs, and
hold it out at arms' length. With both eyes open, focus on
a specific spot in the distance, and center that spot in the
triangle of your fingers. Now hold your arms still and shut
one eye after the other. The spot will be closer to the
center of the triangle when viewed with your dominant
eye.
70
Biology 300 lab manual
Stats lab student data sheet 7
Record the following measurements on yourself. These will be shared (without
your name attached) with the rest of the class, so if you feel any discomfort with
such sharing feel free to omit any or all of the measurements.
Sex
Handedness
Which hand do you
normally write with?
Right index
finger length
Measured in mm from
the basal crease of the
finger to the tip -- see
2D in the figure.) This
is called 2D, because it
is on the second digit,
counting the thumb.
Right ring (4th)
finger length
Measured in mm from
the basal crease of the
finger to the tip-- see
4D in the figure.
Left index finger
length
Same as above, but on
left hand.
Left ring (4th)
finger length
Same as above, but on
left hand.
71
Biology 300 lab manual
Stats lab student data sheet 8
Record the following measurements on yourself. These will be shared (without
your name attached) with the rest of the class, so if you feel any discomfort with
such sharing feel free to omit any or all of the measurements.
Handedness
Hand dexterity Right hand
Hand dexterity Left hand
Which hand do you normally write with?
Take the test at
http://faculty.washington.edu/chudler/java/rldot.html
with each hand. Flip a coin to decide which hand to
do first.
72
Biology 300 lab manual
Tips for using JMP
JMP is a very useful and mostly intuitive program. Here are a few tips to make it
easier to use.
Use the “Analyze” menu
A great deal of what you want JMP to do for you starts in the “Analyze” tab
of the main menu bar. Under this “Distribution” will help you visualize and
describe individual variables, “Fit Y by X” will do most basic statistical
analysis on two variables, and “Multivariate methods” will let you do
correlations (and a lot of other things).
Explore for the red triangles
The analysis windows in JMP often include small red triangles like this .
When you click on these a drop-down menu with many appropriate
choices will appear. These choices let you do further analyses, change the
features of graphs and tables, set key parameters, etc.
Working with columns
A column in JMP represents a variable, or the results of a formula that use
variables from other columns. Column names and properties can be
changed by double-clicking on the column header. You can change
whether JMP thinks of a variable as continuous (i.e., numerical) or
nominal (i.e. categorical) in this way.
The “By” function
Many analyses allow you to look at the data separately “by” a grouping
variable. For example, in the “Distribution” window, if you use a variable
“Sex” in the “By” box, then whatever analysis you ask for will be done
separately for all the males and also all the females.
Finding individuals
The graphs in JMP are cleverly linked to the data file. If you click on a bar
in a histogram, all the individuals that are part of that bar will be selected
in the data file. If you click on a point in a scatterplot, that individual will be
selected. Any selected individuals will then be marked in any graph
associated with that data.
Getting help
JMP has a help button on the right of the main menu bar. It doesn’t always
work smoothly. You may want to Google it with the word JMP added in the
search.
73
Biology 300 lab manual
Copying a graph (or anything else) from JMP
If you want to use a graph for JMP in another file, it helps to know how to
copy and paste from JMP's output. Use the “fat-plus” tool
from the top
of the analysis window.
After you have clicked on the fat-plus, select the part of the window that
you want to copy. Clicking on the section title usually works well. Then
“Copy” from the “File” menu and paste it wherever you like.
Finding P-values
JMP has an idiosyncratic way of giving P-values. Instead of saying
something like P < 0.05, or P = 0.017, JMP will have something that says,
for example, Prob > ChiSq = 0.017, or Prob > t = 0.002, or something
similar. In both cases, this is JMP's way of giving the P-value, P = 0.017 in
the first case and P = 0.002 in the second case.
Please note that JMP is calculating the P-value directly for you, so there is
no need to take the test statistic to a stats table to get the P-value. Of
course, doing this can be a great check on your understanding of both
JMP’s output and how to use the stats tables.
74
Biology 300 lab manual
Writing a Lab Report (compiled by BIOL 300 TAs)
Format
Include a brief descriptive title.
12 point font; Double spaced; 2cm margins; use a legible font.
Figures need to communicate effectively (don’t make them too small).
Don’t go over the page limit (which includes figures!) – points are deducted for
pages over length.
Write in paragraph form.
Label each section with a heading (i.e. Introduction, Methods, etc.).
Tone
Avoid use of slang or casual abbreviations. Read other journal articles to get an
idea of tone used. Note: section lengths given below for Introduction, Methods,
Results and Discussion are suggestions, and not requirements.
Introduction
Introduce the main topics and explain why they are important and why we care
about them.
Define any important concepts/terms that will be referred to in the paper.
Clearly state the question being addressed by your experiment.
Give a brief overview of what you did to address the question.
Mention the biological relevance or usefulness of the study.
Start with broad conceptual information and narrow focus to the point of the study
by the end of the intro. In this course, your intro could be ~1/2 – 1 page.
Methods
Summarize, in paragraph form, what you did.
Choose an appropriate level of detail so that another scientist could repeat your
study, but not so much detail that the reader is overwhelmed or disinterested
(i.e., we don’t need to know who in the group did what, the column titles in JMP,
etc.)
75
Biology 300 lab manual
The methods should be in paragraph form, like the other sections.
Typically in this course the methods might be ~1/2 a page.
Results
Report in paragraph form all findings that are relevant to your research question.
Do not interpret your findings in this section; just state what you found. If
necessary include figures or tables with complete captions.
Every figure and table you include must be referred to in the text.
Figures must include labeled axes (do this in JMP, or by hand) including units
and scale.
Figures should include legends when necessary.
It is standard that figure captions are placed below the figure, whereas table
captions are placed above the table.
Captions should be detailed enough to stand alone. For example: Figure 1. The
difference in circumference between cowrie shells measured by groups A&B.
Mean = 2mm (+/- SE .3 mm)
Text should come before the figures and in the text you should refer to your
figures in numerical order. You should also present your figures in numerical
order.
When you report a mean, you must also report some measure of the uncertainty
in the mean or variance in the population.
In this course, text could be ~1/2 page, and altogether the figures ~ 3/4 to 1 full
page.
Discussion
Discuss the implications of your Results. Explain why you think you found what
you found.
Relate your findings back to your initial question (from the Introduction). Did you
find what you expected? If not, why not?
Briefly summarize any benefits or problems with the design. Mention any relevant
improvements.
Finish with some general conclusions and potentially some ideas for more
interesting follow up experiments In this course, the discussion might take ~1-1.5
pages.
76
Biology 300 lab manual
General
Avoid referencing the TA or saying things like “in the Bio 300 lab, etc.”
There’s no need to talk about the details of how the data were stored (e.g., “we
created columns in JMP labeled A1, A2, etc.”, unless this will make the methods
easier to understand in some way.
In the text, do not describe in detail the figures that are included. Just reference
them and report what they are showing. For example, do not say things like “the
histogram of X has the highest bar at a frequency of 0.4”... this can be easily
seen by someone who looks at the figure.
Every figure and table needs to be referenced in the main text.
Avoid repeating the same word multiple times in a sentence.
Avoid including arbitrary information (e.g., “We labeled group members A1, A2,
B1, B2), unless this somehow makes things clearer later on.
Have someone proofread to check for grammar, tense, spelling, paragraph
structure and writing flow.
77